Right after moving alviergroup.com to its new host, we audited the live site and tested it on an emulated phone on a slow connection. Home page largest content time went from 5.28 to 2.50 seconds, and the pricing page from 3.62 to 1.67 seconds. We also found and fixed technical gaps in canonical tags, redirects, error pages and email records. One item, Microsoft DKIM, is still open.
Why we audited ourselves
We sell websites, so a slow or sloppy site of our own would be hard to explain. On November 18, 2026, we ran two checks on the live site. One was a speed test. The other was an audit of the live pages, which we fetched and parsed for titles, headings, links, structured data, redirects, status codes and headers.
This is our own site and our own measured work, which is why this article says "we". There are no client results here, and no ranking or traffic claims. We did not have Search Console or analytics data to look at, so we cannot say how the site performs in Google.
How we measured speed
We used Chromium with iPhone 13 emulation, a 4x CPU slowdown, and the network limited to 1.6 Mbps with 150 ms of latency. The cache was disabled. Every number below is the median of three runs.
"First content shown" is when the first text or image appears. "Largest content shown" is Largest Contentful Paint, or LCP, which web.dev defines as when the largest visible element renders. web.dev gives 2.5 seconds or less as good, but measured at the 75th percentile of real visits. Our test is one emulated device, so it shows direction, not a pass or fail.
Before and after
| Page | Measure | Before | After |
|---|---|---|---|
| Home | First content shown | 2.85 s | 1.34 s |
| Home | Largest content shown | 5.28 s | 2.50 s |
| Home | Page weight | 383 KB | 361 KB |
| Pricing | Largest content shown | 3.62 s | 1.67 s |
| Pricing | Page weight | 390 KB | 236 KB |
| AFM Screenshots (10 screenshots) | Page weight | 1,651 KB | 1,016 KB |
Our football media site, americanfootball.report, was already at 0.77 seconds to first content under the same test, so we made no speed changes to it. We did not fix what was not broken.
The home page weight barely moved, but its timings roughly halved. That suggests the delay was not mostly about total bytes. It looks more like what the browser had to fetch before it could draw the page.
What we changed
- Fonts moved to our own domain. We stopped loading fonts from Google Fonts and host them ourselves. That removed one third-party font request chain, a set of requests that starts on another server and then has to fetch more files. On a slow, high-latency connection each extra hop adds waiting time.
- Images converted to WebP. The AFM screenshots went from about 1.6 MB to about 0.8 MB. The hero photo went from 329 KB to 128 KB. The logo, which we resized rather than converted, went from 144 KB to 22 KB. web.dev says WebP is usually about 25-35% smaller than JPEG and PNG. Some of our files shrank by more than that.
- Lazy loading for images below the fold. web.dev advises lazy-loading offscreen images but not the first-view image. We followed that on the home page, so the hero loads at once and the images further down wait.
- Hashed filenames and one-year caching. CSS and JavaScript files are renamed using a hash of their content, so a changed file gets a new name. That lets us tell browsers to keep them for a year. Our test disables the cache, so this change does not show in the table. It helps repeat visits.
What the audit found
None of these appear in a speed test. All are quick to fix, and none was a ranking emergency.
- No canonical tags, and www served a duplicate of the main domain. Google's duplicate URL guidance lists different hostnames as a way one page ends up at several addresses, and recommends permanent redirects and
rel="canonical". - The sitemap and every internal link used addresses that redirected. Our host served clean addresses, but the links pointed to
.htmlversions that redirected. The page still worked, but every click took an extra hop and the sitemap listed addresses we did not want indexed. - No structured data on any page. Google says structured data is not required and does not guarantee rich results, and that no special markup is needed for AI Overviews. We still wanted our name, services and founders stated in a machine-readable form that matches the page.
- Nonexistent URLs returned a normal page. A made-up address returned a 200 status instead of a 404. Google calls that a soft 404.
- No social preview tags. A shared link showed no tailored title or image.
- The AFM page had no H1. It used only H2 headings.
- Our email records were incomplete. Our root SPF record did not include Microsoft 365, although our mail runs there, and no DKIM existed for Microsoft. Microsoft says most Microsoft 365 organizations need
include:spf.protection.outlook.comin the SPF record and that SPF alone is not enough, so DKIM and DMARC are also recommended.
What is still open
All of these have been fixed, except Microsoft DKIM, which is still unfinished. Microsoft's documentation describes how it signs outbound mail for a custom domain. Turning it on needs an administrator sign-in to our Microsoft 365 account, and that has not happened yet. Until it does, our mail has an SPF record but no DKIM signature from Microsoft.
We have not measured whether any of this changes search results, and we will not claim it does. What we did measure is that pages load faster on a slow phone and that the site has fewer technical errors. If you want the same checks on your site, our checklist is the place to start, and our website plans include a quality check before launch.
